Rubbish transport vehicle route planning method and system based on multi-source data

Abstract

The invention relates to the technical field of intelligent traffic and smart city waste management, and discloses a path planning method and system for a garbage transport vehicle based on multi-source data, and the method comprises the steps: generating a reference path based on historical collection and transportation data, the method comprises the following steps: performing data cleaning on pre-acquired garbage quantity data, vehicle state data, real-time traffic data, hydroelectric data and fuel consumption data, performing multi-source time sequence feature fusion anomaly detection on a cleaned data set, performing trajectory inversion on the cleaned data set, performing target constraint through a mixed integer programming model, and performing multi-source time sequence feature fusion anomaly detection on the cleaned data set. Then gradient descent parameter optimization is carried out on the initial optimization model, elastic resource scheduling is carried out on an optimized path scheme based on a large event plan library, a path instruction set is generated, dynamic re-optimization evaluation is carried out on the path instruction set based on garbage amount change and traffic condition data fed back in real time, and an updated path is generated; according to the invention, the path planning efficiency of the garbage transport vehicle based on the multi-source data can be improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation and smart city waste management technology, and in particular to a route planning method and system for garbage trucks based on multi-source data. Background Technology

[0002] Current urban waste transport vehicle route planning technologies suffer from significant shortcomings in multi-source data processing and core issue identification capabilities. Existing solutions often employ simple splicing or single-dimensional analysis of multi-source data such as waste volume, vehicle status, and real-time traffic, failing to construct a multi-source temporal feature fusion mechanism. They cannot extract temporal dynamic features through sliding windows and combine them with dimensionality reduction and isolated forest algorithms to achieve accurate anomaly detection. This leads to abnormal data interference in route planning caused by sensor errors and transmission failures. Furthermore, they lack trajectory inversion capabilities based on Kalman filter trajectory smoothing and density clustering, making it difficult to trace waste source areas and providing accurate spatial information on high-frequency waste points for route optimization. This results in a mismatch between route coverage and waste generation needs.

[0003] At the path planning framework level, existing technologies struggle to balance initial optimization with dynamic adaptability. Most solutions rely solely on static algorithms to generate fixed routes, failing to integrate mixed-integer programming models to construct objective functions under multiple constraints. This makes it impossible to achieve cost-minimizing initial path solutions, and they lack dynamic re-optimization mechanisms. They have neither established evaluation models based on real-time changes in waste volume and traffic data nor incorporated large-scale event contingency plans for flexible resource scheduling. When faced with emergencies such as traffic congestion or a surge in waste from large-scale events, path solutions cannot be quickly adjusted, often resulting in a sharp drop in transportation efficiency, resource waste, or task delays. This makes it difficult to meet the dynamic and refined operational needs of urban waste collection and transportation. Therefore, how to solve the mismatch between path coverage and waste generation needs, and how to balance initial optimization with dynamic adaptability, has become an urgent problem to be addressed. Summary of the Invention

[0004] This invention provides a route planning method and system for garbage trucks based on multi-source data to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a route planning method for garbage trucks based on multi-source data, comprising: S1 generates a baseline route based on historical collection and transportation data, and cleans the pre-acquired garbage volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data to generate a cleaned dataset of garbage trucks. S2, perform multi-source temporal feature fusion anomaly detection on the cleaned dataset to generate the anomaly detection result of the garbage truck; S3, Perform trajectory inversion on the cleaned dataset to generate the waste source tracing results of the waste transport vehicle; S4. Based on the cleaned dataset, anomaly detection results, waste tracing results, and the baseline route, apply objective constraints to the preset mixed integer programming model to generate the initial optimization model of the waste transport vehicle. S5, perform gradient descent parameter optimization on the initial optimization model to generate an optimized path scheme for the garbage truck; S6, Based on a large-scale event contingency plan library, perform flexible resource scheduling on the optimized route plan to generate the route instruction set for the garbage truck; S7. Based on real-time feedback of changes in waste volume and traffic conditions, the route instruction set is dynamically re-optimized and evaluated to generate an updated route for the waste transport vehicle.

[0006] In a preferred embodiment, the step of generating a baseline route based on historical collection and transportation data involves cleaning the pre-acquired garbage volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data to generate a cleaned dataset of garbage trucks, including: S201, perform path clustering analysis on historical collection and transportation data to generate baseline routes; S202 integrates the pre-acquired waste volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data to obtain a multi-source raw dataset; S203, imput missing values ​​in the multi-source original dataset to obtain the imputed dataset; S204, perform outlier detection on the filled dataset to obtain a cleaned dataset.

[0007] In a preferred embodiment, the step of performing multi-source temporal feature fusion anomaly detection on the cleaned dataset to generate anomaly detection results for the garbage truck includes: S301, Perform multi-source temporal feature extraction on the cleaned dataset to obtain the multi-source feature vector set of the garbage truck; S302, perform feature fusion on the multi-source feature vector set to obtain the fused feature matrix of the garbage truck; S303, perform anomaly detection analysis on the fused feature matrix to obtain the anomaly score of the garbage truck; S304, perform a threshold judgment on the abnormal score to obtain the abnormal detection result of the garbage truck.

[0008] In a preferred embodiment, the step of performing trajectory inversion on the cleaned dataset to generate the waste tracing results for the waste transport vehicle includes: S401, Extract the location coordinate sequence and timestamp sequence from the cleaned dataset to obtain the original trajectory data of the garbage truck; S402, Perform trajectory smoothing processing on the original trajectory data to obtain a smooth trajectory sequence of the garbage truck; S403, Perform a stop point detection operation on the smooth trajectory sequence to obtain the stop point set of the garbage truck; S404. Using a density clustering algorithm, the set of stopping points is analyzed to determine the waste source region, thereby obtaining the waste source tracing results of the waste transport vehicle.

[0009] In a preferred embodiment, the step of applying objective constraints to a preset mixed-integer programming model based on the cleaned dataset, anomaly detection results, waste source tracing results, and the baseline route to generate an initial optimization model for the waste transport vehicle includes: S501, Based on the cleaned dataset, anomaly detection results, waste tracing results, and the baseline route, extract the optimized parameter set of the waste transport vehicle to obtain the model input parameters; S502, Based on the model input parameters, construct the objective function and constraint set of the mixed integer programming model; S503, apply objective constraints to the objective function and constraint set to obtain the initial optimization model of the garbage truck.

[0010] In a preferred embodiment, constructing the objective function and constraint set of the mixed integer programming model based on the model input parameters includes: S601, Based on the model input parameters, define the set of decision variables for the mixed integer programming model; S602, Based on the set of decision variables and model input parameters, construct the objective function of the garbage truck; S603, Based on the model input parameters and business requirements, define the constraint set for the garbage truck.

[0011] In a preferred embodiment, the step of optimizing the initial optimization model using gradient descent parameters to generate an optimized path scheme for the garbage truck includes: S701, Perform parameter initialization on the initial optimization model to obtain an initial parameter set; S702, perform gradient descent iterative optimization on the initial parameter set to obtain the optimized parameter set; S703, perform path reconstruction on the optimized parameter set to obtain the optimized path scheme for the garbage truck.

[0012] In a preferred embodiment, the step of flexibly scheduling resources for the optimized route plan based on a large event contingency plan library to generate the route instruction set for the garbage truck includes: S801, perform event matching between the optimized path scheme and the large event contingency plan library to obtain a matching contingency plan set; S802, perform resource scheduling optimization on the matching plan set to obtain an elastic scheduling scheme; S803, perform instruction generation and conversion on the elastic scheduling scheme to generate the route instruction set for the garbage truck.

[0013] In a preferred embodiment, the dynamic re-optimization and evaluation of the route instruction set based on real-time feedback of waste volume changes and traffic condition data to generate an updated route for the waste transport vehicle includes: S901 integrates real-time feedback data on changes in waste volume and traffic conditions to obtain a dynamically updated dataset. S902, the dynamically updated dataset and route instruction set are re-optimized and evaluated to obtain the evaluation results; S903, adjust the route based on the evaluation results to generate an updated route for the garbage truck.

[0014] To address the above problems, the present invention also provides a route planning system for garbage trucks based on multi-source data, the device comprising: The data cleaning module generates a baseline route based on historical collection and transportation data. It cleans the pre-acquired garbage volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data to generate a cleaned dataset of garbage trucks. The anomaly detection module performs multi-source temporal feature fusion anomaly detection on the cleaned dataset to generate anomaly detection results for the garbage truck. The waste tracing module performs trajectory inversion on the cleaned dataset to generate the waste tracing results for the waste transport vehicle; The objective constraint module, based on the cleaned dataset, anomaly detection results, waste tracing results, and the baseline route, applies objective constraints to a preset mixed integer programming model to generate an initial optimization model for the waste transport vehicle. The parameter optimization module performs gradient descent parameter optimization on the initial optimization model to generate an optimized path scheme for the garbage truck. The elastic resource scheduling module, based on a large event contingency plan library, performs elastic resource scheduling on the optimized route plan and generates the route instruction set for the garbage truck. The dynamic re-optimization module performs dynamic re-optimization evaluation on the route instruction set based on real-time feedback of changes in waste volume and traffic conditions, and generates updated routes for the waste transport vehicles.

[0015] This invention significantly improves data processing accuracy and problem identification capabilities by constructing a waste tracing and anomaly detection mechanism based on multi-source temporal feature fusion and trajectory inversion. On the one hand, it uses a sliding window to extract temporal features such as mean and variance from the cleaned multi-source data, and combines principal component analysis dimensionality reduction and isolated forest algorithm to achieve anomaly detection. This can accurately identify abnormal states such as sensor errors and vehicle malfunctions, effectively avoiding interference from abnormal data on path planning. On the other hand, it smooths the trajectory through Kalman filtering, detects stopping points through speed thresholds, and completes waste tracing through density clustering. This can accurately locate high-frequency waste generation areas, providing targeted spatial basis for path planning. It solves the problems of poor path targeting caused by insufficient data fusion, lagging anomaly identification, and lack of tracing capabilities in traditional solutions, ensuring that path coverage is highly matched with waste collection and transportation needs.

[0016] This invention integrates a flexible path planning framework combining mixed-integer programming and dynamic re-optimization, significantly improving the optimality and adaptability of path solutions. In the initial optimization stage, a mixed-integer programming model with constraints such as path continuity, vehicle capacity, and time window is constructed based on parameters extracted from multi-source data. Combined with gradient descent parameter optimization, it can generate an initial path with the minimum total cost. In the dynamic adjustment stage, it relies on a large-scale event contingency plan library to achieve path-to-contingency plan matching and flexible resource scheduling. At the same time, it performs dynamic re-optimization evaluation based on real-time waste volume and traffic data to quickly generate updated routes. This framework solves the shortcomings of traditional static planning in dealing with emergencies and rigid resource scheduling. It can flexibly adapt to scenarios such as traffic congestion, sudden changes in waste volume, and large-scale events, significantly improving waste transportation efficiency, reducing resource waste, and meeting the needs of refined and dynamic operation of urban waste collection and transportation. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a route planning method for garbage trucks based on multi-source data, provided in an embodiment of the present invention. Figure 2 This is a functional block diagram of a route planning system for a garbage truck based on multi-source data, provided in an embodiment of the present invention. The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] This application provides a route planning method for garbage trucks based on multi-source data. The execution entity of the route planning method for garbage trucks based on multi-source data includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the route planning method for garbage trucks based on multi-source data can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0020] Reference Figure 1 The diagram shown is a flowchart illustrating a route planning method for garbage trucks based on multi-source data, according to an embodiment of the present invention. In this embodiment, the route planning method for garbage trucks based on multi-source data includes: S1 generates a baseline route based on historical collection and transportation data, and cleans the pre-acquired garbage volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data to generate a cleaned dataset of garbage trucks. In this embodiment of the invention, the step of generating a baseline route based on historical collection and transportation data involves cleaning the pre-acquired garbage volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data to generate a cleaned dataset of garbage trucks, including: S201, perform path clustering analysis on historical collection and transportation data to generate baseline routes; S202 integrates the pre-acquired waste volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data to obtain a multi-source raw dataset; S203, imput missing values ​​in the multi-source original dataset to obtain the imputed dataset; S204, perform outlier detection on the filled dataset to obtain a cleaned dataset.

[0021] It should be noted that the path clustering analysis is based on the historical collection point set, which is divided into point clusters to generate a cluster center point set. Then, based on the cluster center point set, the nearest neighbor algorithm is used to generate the baseline route, dividing the historical collection point set into k clusters. The center point of each cluster is calculated as the average coordinate of all points in the cluster. The goal of the algorithm is to minimize the sum of squared Euclidean distances between all points in the cluster and the cluster center point. The value of k is determined by optimization based on historical data. Essentially, it reduces the complexity of route planning by iteratively converging the collection points into representative clusters.

[0022] It should be noted that data integration is based on pre-acquired waste volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data. Multi-source data is integrated through a data fusion algorithm to generate a multi-source raw dataset. The data fusion algorithm uses timestamp alignment and spatial index matching to unify data from different sources into the same spatiotemporal framework and uses hash table join operations to achieve data association.

[0023] It should be noted that the multi-source raw dataset includes garbage volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data. Among them, garbage volume data represents the garbage volume value at each time point, vehicle status data represents the vehicle status vector, real-time traffic data represents the traffic condition value, water and electricity data represents the water and electricity consumption value, and fuel consumption data represents the fuel consumption value. It is used to provide comprehensive information on the operating environment of garbage trucks, and to characterize the dynamic factors in the transportation process through multi-dimensional data. This provides basic data support for anomaly detection, trajectory inversion, and optimization model construction, ensuring the comprehensiveness and accuracy of route planning.

[0024] It should be noted that missing value imputation is based on a multi-source original dataset. Data is filled by interpolation algorithm to generate the imputed dataset. The interpolation algorithm uses linear interpolation to estimate the missing time point using the value of the nearest non-missing point before and after it. The imputed value is calculated as the value of the previous non-missing point plus the proportion of the time interval multiplied by the difference between the next non-missing point and the previous non-missing point.

[0025] Furthermore, missing value imputation essentially restores missing information by assuming data continuity, improves the integrity of the dataset, avoids bias in subsequent analysis due to missing data, ensures the continuity and consistency of time series data, provides complete input for feature extraction and model training, and enhances the reliability of route planning.

[0026] It should be noted that the outlier detection operation is based on the imputed dataset. Noise is identified and corrected using a statistical detection algorithm to generate a cleaned dataset. The statistical detection algorithm uses the Z-score method to calculate the Z-score value of each data point. The Z-score value is the absolute difference between the data point value and the mean of the data sequence divided by the standard deviation of the data sequence. If the Z-score value exceeds a preset threshold, such as 2.5, the data point is regarded as an outlier and corrected.

[0027] Furthermore, outlier detection essentially involves identifying data points that deviate from the normal pattern using a distribution model, eliminating noise caused by sensor errors or transmission failures, improving data quality, reducing the interference of outlier data on subsequent steps, and ensuring the accuracy of data analysis and the stability of the model.

[0028] It should be noted that the cleaned dataset includes cleaned waste volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data. These data have undergone missing value imputation and outlier detection processing, ensuring their completeness, consistency, and reliability.

[0029] Furthermore, the cleaned dataset serves as input for subsequent multi-source temporal feature fusion anomaly detection, trajectory inversion, and mixed-integer programming model construction. High-quality data supports the entire route planning process, improving the operational efficiency and resource utilization of garbage trucks.

[0030] S2, perform multi-source temporal feature fusion anomaly detection on the cleaned dataset to generate the anomaly detection result of the garbage truck; In this embodiment of the invention, the step of performing multi-source temporal feature fusion anomaly detection on the cleaned dataset to generate the anomaly detection result for the garbage truck includes: S301, Perform multi-source temporal feature extraction on the cleaned dataset to obtain the multi-source feature vector set of the garbage truck; S302, perform feature fusion on the multi-source feature vector set to obtain the fused feature matrix of the garbage truck; S303, perform anomaly detection analysis on the fused feature matrix to obtain the anomaly score of the garbage truck; S304, perform a threshold judgment on the abnormal score to obtain the abnormal detection result of the garbage truck.

[0031] It should be noted that multi-source time series feature extraction is based on time series data from the cleaned dataset. Feature calculation is performed using the sliding window algorithm to generate a multi-source feature vector set. The time series data is divided into fixed-length windows, and statistical features, including mean, variance, maximum and minimum values, are calculated for each window. Specifically, the time series of the cleaned dataset contains data values ​​from time point 1 to time point 2, where each data value corresponds to a time point. The sliding window has a fixed length and sliding step size, and each window covers consecutive data points. For example, the i-th window covers data points i to data points i plus the sliding window length minus 1. For each window, a feature vector is calculated, which includes the window mean, window variance, maximum and minimum values ​​within the window. The window mean is the arithmetic mean of all data values ​​within the window, the window variance is the average of the squared differences between each data value within the window and the mean, and the maximum and minimum values ​​within the window are the maximum and minimum values ​​of the data values ​​within the window, respectively. The multi-source feature vector set is composed of the feature vectors of all windows. The number of windows is calculated by the total number of time points, the sliding window length, and the step size, specifically by subtracting the sliding window length from the total number of time points, dividing by the sliding step size, rounding down, and adding 1.

[0032] Furthermore, the sliding window length and step size are optimized based on the data sampling frequency and business requirements. Essentially, this captures the time-series dynamic patterns through local statistics, reduces the impact of noise, and provides a stable input for subsequent fusion.

[0033] Furthermore, the multi-source feature vector set contains features of waste volume, vehicle status, real-time traffic, water and electricity consumption, and fuel consumption data, which are used to comprehensively characterize the operating status of transport vehicles.

[0034] It should be noted that the multi-source feature vector set includes garbage volume data features, vehicle status data features, real-time traffic data features, water and electricity data features, and fuel consumption data features extracted from the cleaned dataset. Each feature vector corresponds to a statistical feature of a sliding window, including the window mean, variance, maximum value, and minimum value. These feature vectors together constitute the multi-source feature vector set, which is used to characterize the operating status of garbage trucks in different time windows. It is used to capture the dynamic changes of time series data through multi-dimensional statistical features, provide input for feature fusion, support subsequent anomaly detection and analysis, and ensure that route planning is based on comprehensive operating status information.

[0035] It should be noted that feature fusion is based on a multi-source feature vector set. Principal component analysis (PCA) is used to perform dimensionality reduction and integration to generate a fused feature matrix. Specifically, the analysis algorithm calculates the covariance matrix of the multi-source feature vector set, performs eigenvalue decomposition, selects the eigenvectors corresponding to the k largest eigenvalues ​​to form a projection matrix, and projects the original feature vectors into a lower-dimensional space to form the fused feature matrix. Feature fusion essentially eliminates feature redundancy through linear transformation, retains the main variation information of the data, and improves computational efficiency by using dimensionality reduction techniques, providing a concise and effective input for anomaly detection.

[0036] It should be noted that the fusion feature matrix contains multi-source feature vectors after dimensionality reduction. Each row vector represents the fusion feature of a window, and the column vectors represent the feature values ​​in different principal component directions. The matrix dimension is the number of windows multiplied by the number of principal components. It is used as input for anomaly detection analysis. By using low-dimensional representation, key information of the data is preserved, the model complexity is simplified, and the efficiency of anomaly detection is improved.

[0037] It should be noted that the anomaly detection analysis is based on the fusion feature matrix and uses the isolation forest algorithm to identify anomalies and generate anomaly scores. The isolation forest algorithm constructs multiple random trees and calculates the path length of each sample by randomly splitting the data. The shorter the path length, the easier it is for the sample to be isolated. The anomaly score is calculated as an exponential function based on the average path length. The higher the score, the greater the probability of an anomaly.

[0038] Furthermore, anomaly detection analysis essentially simulates the ease of isolating anomalies through unsupervised learning, improves detection stability through ensemble learning, adapts to the complex distribution of multi-source data, quantifies the degree of anomaly in each window, and automatically identifies anomaly patterns in the running state through algorithms, providing a basis for threshold judgment.

[0039] It should be noted that the anomaly score is a continuous value representing the degree of anomaly in each window. It is calculated based on the average path length in the isolated forest algorithm. The score ranges from 0 to 1. The closer the score is to 1, the more highly anomaly it is, and the closer it is to 0, the more normal it is. It represents the degree to which data points deviate from the normal pattern through mathematical modeling, and is used to measure the stability of the garbage truck's operating status, providing a reference for dynamic path adjustment.

[0040] It should be noted that the threshold determination is based on the anomaly score sequence. A dynamic thresholding algorithm is used for binary classification to generate anomaly detection results. The dynamic thresholding algorithm calculates the mean and standard deviation of the anomaly score sequence and sets the threshold. In the formula, The mean of outlier scores. Standard deviation, This is for adjusting the coefficient.

[0041] Furthermore, the adjustment coefficient is set to 2.0. If the abnormal score is greater than T, it is marked as abnormal; otherwise, it is normal.

[0042] It should be noted that the anomaly detection results contain a binary sequence, with each element corresponding to a window. 1 indicates an anomaly and 0 indicates normal operation. The sequence length is consistent with the number of windows. This sequence is used to identify abnormal time periods or areas during the operation of garbage trucks, providing input for a mixed-integer programming model to avoid the impact of anomalies, optimize route selection, and improve transportation safety and efficiency.

[0043] S3, Perform trajectory inversion on the cleaned dataset to generate the waste source tracing results of the waste transport vehicle; In this embodiment of the invention, the step of performing trajectory inversion on the cleaned dataset to generate the waste source tracing results of the waste transport vehicle includes: S401, Extract the location coordinate sequence and timestamp sequence from the cleaned dataset to obtain the original trajectory data of the garbage truck; S402, Perform trajectory smoothing processing on the original trajectory data to obtain a smooth trajectory sequence of the garbage truck; S403, Perform a stop point detection operation on the smooth trajectory sequence to obtain the stop point set of the garbage truck; S404. Using a density clustering algorithm, the set of stopping points is analyzed to determine the waste source region, thereby obtaining the waste source tracing results of the waste transport vehicle.

[0044] It should be noted that the original trajectory data is the location coordinate sequence and timestamp sequence extracted from the cleaned dataset. The location coordinate sequence represents the latitude and longitude coordinates of the garbage truck during transportation, and the timestamp sequence represents the collection time of the corresponding coordinates. The purpose of the original trajectory data is to provide the original path information of the vehicle's movement, serving as the starting point for trajectory inversion and supporting subsequent trajectory smoothing and stop point detection to reconstruct the actual transportation path.

[0045] It should be noted that trajectory smoothing is based on the original trajectory data. Noise is removed using the Kalman filter algorithm to generate a smooth trajectory sequence. The position sequence of the original trajectory data is used as input, where each position point consists of longitude and latitude coordinates. The state vector of the Kalman filter includes position coordinates and velocity components. The state transition equation is based on the assumption of uniform motion and uses a state transition matrix to describe the change of state over time. The elements of the state transition matrix are determined by the time interval. Both process noise and observation noise are assumed to follow a Gaussian distribution with a mean of zero. The observation equation maps the state vector to the observation values. The posterior state estimate is calculated iteratively through prediction and update steps to generate a smooth trajectory sequence, where the smoothed coordinates are obtained from the state estimate.

[0046] Furthermore, the physical essence of the Kalman filter algorithm is to fuse historical trajectories and current observations through a dynamic system model, and use recursive estimation to reduce GPS noise and jitter, thereby improving trajectory accuracy. It is suitable for optimizing the positioning data of garbage trucks in complex environments.

[0047] It should be noted that the smoothed trajectory sequence is a trajectory sequence generated by removing noise from the original trajectory data using the Kalman filter algorithm. It includes smoothed position coordinates and timestamps, which is used to eliminate GPS positioning errors and jitter, improve the accuracy and continuity of trajectory data, provide reliable input for stop point detection, and reduce the impact of noise on waste source tracing analysis.

[0048] It should be noted that the dwell point detection operation is based on a smooth trajectory sequence. Dwell points are identified through a speed threshold algorithm, generating a dwell point set. The speed threshold algorithm calculates the instantaneous speed of each consecutive pair of points in the smooth trajectory sequence. The instantaneous speed is the Euclidean distance between the point pairs divided by the time interval. If the instantaneous speed is less than a preset speed threshold, the point is marked as a potential dwell point. Consecutive potential dwell points are merged. If the duration of consecutive points exceeds a preset time threshold, the average coordinates of these points are calculated as the dwell point. The dwell point set consists of all dwell points that meet the conditions.

[0049] It should be noted that the set of dwell points is a collection of dwell points identified from the smooth trajectory sequence using a speed threshold algorithm. Each dwell point represents the time a garbage truck stayed at a certain location. This is used to identify the key positions of the vehicle in actual operation, provide input for density clustering analysis, and support the identification of garbage source areas.

[0050] It should be noted that the density clustering algorithm is based on the set of dwell points to perform spatial clustering and generate waste source areas. The algorithm defines two parameters: neighborhood radius and minimum number of points. For each dwell point, the number of points in its neighborhood is calculated. If the number of points is greater than or equal to the minimum number of points, the point is marked as a core point, and the density-connected points are recursively expanded to form clusters, and finally a set of clusters is generated. Each cluster represents a waste source area.

[0051] Furthermore, density clustering algorithms essentially discover spatial clustering patterns through density reachability, utilize parallel computing to process large-scale point sets, automatically identify waste hotspots, do not require a preset number of clusters, adapt to dynamic urban environments, and are used to identify frequently collected waste areas through spatial clustering analysis of dwell points, providing a data foundation for waste source tracing and supporting priority processing in path optimization.

[0052] It should be noted that the waste source tracing result is a set of waste source areas generated by a density clustering algorithm. Each area is represented by the coordinates of the cluster center point, which identifies the operation points where waste transport vehicles frequently stop. This data serves as input to a mixed-integer programming model to optimize route planning, prioritize services for high-density waste areas, and improve transportation efficiency and resource utilization.

[0053] S4. Based on the cleaned dataset, anomaly detection results, waste tracing results, and the baseline route, apply objective constraints to the preset mixed integer programming model to generate the initial optimization model of the waste transport vehicle. In this embodiment of the invention, the step of applying objective constraints to a preset mixed-integer programming model based on the cleaned dataset, anomaly detection results, waste tracing results, and the baseline route to generate an initial optimization model for the waste transport vehicle includes: S501, Based on the cleaned dataset, anomaly detection results, waste tracing results, and the baseline route, extract the optimized parameter set of the waste transport vehicle to obtain the model input parameters; S502, Based on the model input parameters, construct the objective function and constraint set of the mixed integer programming model; S503, apply objective constraints to the objective function and constraint set to obtain the initial optimization model of the garbage truck.

[0054] It should be noted that the extraction of the optimized parameter set of the garbage truck is based on the cleaned dataset, anomaly detection results, garbage source tracing results, and baseline route. Key parameters are calculated through a parameter fusion algorithm to generate model input parameters, as follows: The cleaned dataset contains garbage volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data. Each dataset contains multiple data points, where the garbage volume data point represents the garbage volume value at each time point, the vehicle status data point represents the vehicle status vector, the real-time traffic data point represents the traffic condition value, the water and electricity data point represents the water and electricity consumption value, and the fuel consumption data point represents the fuel consumption value.

[0055] Furthermore, the anomaly detection results are binary sequences. Each element indicates whether the corresponding detection point is abnormal, with 1 indicating anomaly and 0 indicating normal. The waste source tracing results are a set of source regions, where each region is represented by its center coordinates, and the baseline route is an ordered sequence of path points, with each path point represented by coordinates.

[0056] Furthermore, the parameter fusion algorithm calculates the optimization parameter set, including average waste volume, anomaly weight, source area density, path length, and comprehensive fuel consumption coefficient. The average waste volume is the arithmetic mean of all waste volume data, the anomaly weight is the proportion of anomaly points in the anomaly detection results, the source area density is the ratio of the number of source areas to the total area, the path length is the sum of the Euclidean distances between consecutive path points in the baseline route, and the comprehensive fuel consumption coefficient is the arithmetic mean of all fuel consumption data.

[0057] Furthermore, anomaly weights are used to adjust the model's avoidance of anomaly regions, source region density is used to prioritize high-density garbage areas, and path length is used as a benchmark to optimize travel distance.

[0058] It should be noted that the model input parameters are a set of optimized parameters extracted from the cleaned dataset, anomaly detection results, waste source tracing results, and baseline routes through a parameter fusion algorithm. These parameters include average waste volume, anomaly weight, source area density, path length, and comprehensive fuel consumption coefficient. These parameters quantify the key influencing factors of waste transport vehicle operation and serve as input to the mixed-integer programming model. They provide standardized data for constructing the objective function and constraints, support path optimization decisions, and improve the accuracy and efficiency of model calculations.

[0059] It should be noted that the objective function and constraint set for constructing the mixed-integer programming model are based on the model input parameters. Mathematical modeling is performed using a model definition algorithm to generate the objective function and constraint set. The objective function is defined as minimizing the total cost, as shown in the following mathematical expression: In the formula, Indicates the number of path points. and Indicates the path point index. For path cost, This represents a binary path selection variable.

[0060] Furthermore, the number of waypoints is derived from the waypoint sequence in the baseline route, and the waypoint index ranges from 1 to... , representing the start and end points of the path, respectively; path cost, dependent on model input parameters; binary path selection variable, if the garbage truck starts from the path point... Drive to waypoint ,but ,otherwise The constraint set includes path continuity constraints, vehicle capacity constraints, time window constraints, and resource constraints, based on model input parameters and business requirements.

[0061] Furthermore, constructing the objective function and constraint set essentially involves combining linear programming with integer variables to simulate path selection and resource constraints, and using mathematical optimization theory to ensure the feasibility and optimality of the model.

[0062] Furthermore, a set of objective functions and constraints is constructed to form a complete optimization problem framework, transforming parameters derived from multi-source data into a solvable mathematical model, providing a foundation for subsequent objective constraints and path optimization, and improving the robustness and practicality of route planning.

[0063] It should be noted that the objective constraint is based on the objective function and the set of constraint conditions. The model is instantiated through the constraint injection algorithm to generate an initial optimization model. The constraint injection algorithm injects the objective function and the set of constraint conditions into a pre-defined mixed integer programming model framework. For example, a linear programming solver is used to verify the feasibility of the constraints and adjust infeasible constraints to ensure that the model is solvable.

[0064] Furthermore, objective constraints are used to transform abstract objective functions and constraints into concrete, executable optimization models, providing input for gradient descent parameter optimization and ensuring that the path scheme meets actual operational needs.

[0065] It should be noted that the initial optimization model is a mixed-integer programming model instance generated through objective constraint operations. It includes an objective function, a set of constraints, and a set of decision variables. The role of the initial optimization model is to serve as the basis for subsequent gradient descent parameter optimization, providing a complete data representation for iteratively solving for the optimal path, supporting dynamic route adjustments for garbage trucks, and improving transportation efficiency and resource utilization.

[0066] In this embodiment of the invention, constructing the objective function and constraint set of the mixed integer programming model based on the model input parameters includes: S601, Based on the model input parameters, define the set of decision variables for the mixed integer programming model; S602, Based on the set of decision variables and model input parameters, construct the objective function of the garbage truck; S603, Based on the model input parameters and business requirements, define the constraint set for the garbage truck.

[0067] It should be noted that the decision variable set for the mixed-integer programming model is generated by setting discrete and continuous variables through a variable definition algorithm based on the model input parameters. Specifically, the variable definition algorithm is based on the set of path points, where the number of path points comes from the path point sequence of the baseline route. The decision variables include binary path selection variables, which indicate whether the garbage truck travels from one path point to another. If so, the value is 1, and otherwise it is 0. Auxiliary variables include continuous variables, such as the time to arrive at each path point and the amount of garbage at that path point.

[0068] Furthermore, the physical essence of the decision variable set is defined by simulating the discreteness of path selection through binary variables and using continuous variables to track time and resource consumption, providing a mathematical basis for the optimization model and ensuring that the model can handle real-world path decision problems.

[0069] Furthermore, a set of decision variables is defined to establish the core structure of the mixed-integer programming model. By mapping variable path selection and state information, the objective function and constraints can be constructed, thereby improving the solvability and practicality of the model.

[0070] It should be noted that the objective function is constructed based on the set of decision variables and model input parameters. The objective function is generated by executing the function definition through a cost minimization algorithm. Specifically, the cost minimization algorithm defines the objective function as minimizing the total cost, which is calculated as the sum of the costs between all path points. The cost of each path point pair consists of a weighted sum of distance cost, anomaly avoidance cost, and source region priority cost. The distance cost is calculated based on the Euclidean distance between path points, the anomaly avoidance cost considers anomaly weights and anomaly indicators, and the source region priority cost considers source region density and influence factors. All costs are balanced through weight coefficients.

[0071] Furthermore, constructing the objective function essentially involves linearly combining multiple influencing factors to integrate distance cost, abnormal avoidance cost, and source region priority cost into a single objective. Optimization theory is used to guide path selection, achieving efficient resource allocation. This is used to quantify the economy and efficiency of path planning, providing optimization direction for mixed-integer programming models, ensuring that the path scheme minimizes the total cost, and improving the operational efficiency of garbage trucks.

[0072] It should be noted that defining the constraint set is based on the model input parameters and business requirements. The constraint modeling algorithm executes the condition settings to generate the constraint set, and the mathematical expression of the constraint modeling algorithm is as follows: Path continuity constraints: Vehicle capacity constraints: Time window constraints: Resource constraints: In the formula, path point The amount of garbage, The maximum garbage capacity of the vehicle. To reach the waypoint Time, For the maximum allowed time, path point Water and electricity consumption value, Fuel consumption value This is the maximum resource limit.

[0073] Furthermore, for all Ensure that each waypoint is visited only once. The amount of waste is derived from the cleaned dataset, the maximum allowable time is calibrated based on real-time traffic data, and the fuel consumption value is derived from the cleaned dataset.

[0074] Furthermore, a set of constraints is defined to impose real-world constraints, preventing path breakage, overload, timeout, and resource overrun, ensuring the executability of the path solution in business scenarios, and improving the model's real-world applicability and reliability.

[0075] S5, perform gradient descent parameter optimization on the initial optimization model to generate an optimized path scheme for the garbage truck; In this embodiment of the invention, the step of optimizing the gradient descent parameters of the initial optimization model to generate an optimized path scheme for the garbage truck includes: S701, Perform parameter initialization on the initial optimization model to obtain an initial parameter set; S702, perform gradient descent iterative optimization on the initial parameter set to obtain the optimized parameter set; S703, perform path reconstruction on the optimized parameter set to obtain the optimized path scheme for the garbage truck.

[0076] It should be noted that the parameter initialization operation is based on the initial optimization model. The parameters are set through a random initialization algorithm to generate an initial parameter set. The initial values ​​are sampled from a uniform distribution, as follows: The initial optimization model contains a continuous parameter vector, where each parameter represents an optimizable parameter, such as the weight coefficient in the objective function. The number of parameters is K. The algorithm generates an initial value for each parameter, which is sampled from a uniform distribution on a preset interval, such as a lower limit of 0 and an upper limit of 1. The initial value represents the starting point of the parameter.

[0077] Furthermore, the random initialization algorithm essentially avoids getting stuck in local optima through randomness, and uses probability distributions to provide diverse starting points for gradient descent, thereby increasing the probability of global convergence.

[0078] It should be noted that the gradient descent algorithm optimizes parameters based on the initial parameter set to generate an optimized parameter set. The objective function is defined as a function of the total cost of the initial optimized model as the parameters, where the total cost is the sum of path costs, and the path costs depend on the parameter vector. The algorithm updates the parameters in each iteration by subtracting the product of the learning rate and the gradient vector of the objective function at the current parameter vector from the current parameter vector. Here, the current parameter vector represents the parameter values ​​of the iteration step, the learning rate is a positive scalar used to control the update step size, and the gradient vector is a vector composed of the partial derivatives of the objective function with respect to each parameter. The iteration terminates when the Euclidean norm of the gradient vector is less than a preset gradient threshold or when the maximum number of iterations is reached. The optimized parameter set is the parameter vector at the termination iteration step.

[0079] Furthermore, the gradient descent algorithm essentially adjusts the parameters step by step in the negative gradient direction to minimize the objective function, and uses local linear approximation to guide the parameters to converge toward the optimal solution, making it suitable for optimization in continuous parameter space.

[0080] Furthermore, the learning rate is a hyperparameter in the gradient descent algorithm, representing a positive scalar of the parameter update step size. It controls the magnitude of the parameter vector adjustment along the negative gradient direction in each iteration. The learning rate balances the convergence speed and stability of the optimization process: an excessively large learning rate may lead to optimization oscillations or divergence, while an excessively small learning rate will slow down convergence. In garbage truck route planning, adaptively adjusting the learning rate ensures that gradient descent efficiently converges to the minimum of the objective function, thereby improving path optimization efficiency and model robustness.

[0081] It should be noted that the optimization parameter set is a parameter vector generated through gradient descent iterative optimization. It includes continuously optimized weight coefficients and data-driven parameters, specifically: distance weights, anomaly avoidance weights, source region priority weights, and comprehensive cost coefficients in the objective function. These parameters are derived from the objective function definition of the initial optimization model and are used as inputs for path reconstruction. By minimizing the total cost, the path scheme is ensured to achieve an optimal balance among factors such as distance, anomaly avoidance, and region priority, thereby improving the operating efficiency and resource utilization of garbage trucks and supporting the model's adaptation to dynamic data changes.

[0082] It should be noted that path reconstruction is a method for generating specific path schemes based on an optimized parameter set. This is achieved through a mixed-integer programming solver. The method takes the optimized parameter set as input and re-solves the mixed-integer programming model's decision variable set, including binary path selection variables and continuous state variables, thereby generating an ordered sequence of path points. Specifically: First, the optimized parameter set is substituted into the objective function to minimize the total cost, which is a weighted sum of distance cost, out-of-way avoidance cost, and source region priority cost. Second, under the constraint set, the solver calculates the optimal path selection variable values. Finally, a smooth path sequence is constructed based on these variable values. The role of path reconstruction is to transform abstract optimization parameters into practically feasible path schemes, ensuring that the paths meet business constraints and minimize the total cost, thereby improving the dynamic route adjustment capability and operational efficiency of garbage trucks.

[0083] It should be noted that the optimized route scheme is generated through route reconstruction, which produces an ordered sequence of path points and related metadata, including: the path point coordinate sequence, the estimated arrival time of each path point, the amount of waste, the resource consumption value and water and electricity consumption, and the total route cost. These data are derived from the solution output of the mixed integer programming model. The path point sequence is generated based on the connection relationship where the binary path selection variable is equal to 1. It serves as the actual driving instruction for the garbage truck, guiding the vehicle to complete the collection and transportation task efficiently and safely. By integrating multi-source data, the optimality of the route in terms of cost, time and environmental factors is ensured, thereby improving transportation efficiency, reducing resource waste, and supporting subsequent flexible resource scheduling and dynamic re-optimization.

[0084] S6, Based on a large-scale event contingency plan library, perform flexible resource scheduling on the optimized route plan to generate the route instruction set for the garbage truck; In this embodiment of the invention, the step of performing elastic resource scheduling on the optimized route scheme based on a large event contingency plan library to generate the route instruction set for the garbage truck includes: S801, perform event matching between the optimized path scheme and the large event contingency plan library to obtain a matching contingency plan set; S802, perform resource scheduling optimization on the matching plan set to obtain an elastic scheduling scheme; S803, perform instruction generation and conversion on the elastic scheduling scheme to generate the route instruction set for the garbage truck.

[0085] It should be noted that the event matching operation is based on optimized path schemes and a large event contingency plan library. A similarity calculation algorithm is used to perform contingency plan matching to generate a set of matching contingency plans. The optimized path scheme consists of multiple path point coordinates, and the large event contingency plan library contains multiple event contingency plans. Each contingency plan includes attributes such as event type, affected area, and resource requirements. For each event contingency plan, its similarity with the optimized path scheme is calculated. The similarity is the number of intersection elements of the optimized path scheme's path point set and the contingency plan's affected area path point set divided by the number of union elements. If the similarity is greater than a preset similarity threshold, the contingency plan is included in the set of matching contingency plans.

[0086] Furthermore, similarity calculation algorithms essentially measure the degree of overlap between paths and event regions through similarity, utilize spatial analysis to identify potentially impactful events, provide targeted input for resource scheduling, and improve the accuracy of event response.

[0087] Furthermore, the similarity threshold is a positive scalar value set through historical data analysis and business requirement optimization. It is used to determine whether a contingency plan should be included in the matching contingency plan set during event matching operations. The specific setting method includes: calculating the threshold through offline training based on historical event data and path overlap statistics, and dynamically optimizing it through online adjustment. The threshold is set to 0.6 to balance the risks of false positives and false negatives. The role of the similarity threshold is to ensure that only event contingency plans that are highly related to the optimized path plan are selected, thereby improving the accuracy and efficiency of resource scheduling and reducing unnecessary resource waste.

[0088] It should be noted that the matching plan set is a subset of event plans generated through event matching operations. It includes: the event type, the coordinate sequence of the affected area, resource demand data, and priority flags for each matching plan. These data are derived from the filtered output of a large event plan library. The role of the matching plan set is to serve as input for resource scheduling optimization, identify large events that may affect the route of garbage trucks, support flexible resource allocation by identifying potential risks in advance, and ensure that route planning remains stable and adaptable when events occur.

[0089] It should be noted that resource scheduling optimization is a method based on a matching plan set. It uses a linear programming algorithm to perform resource reallocation to generate a flexible scheduling scheme. The optimization method takes minimizing the total resource cost as the objective function, while considering resource demand constraints, vehicle capacity, time windows, and personnel limitations in the matching plan set. The algorithm uses the simplex method to solve for the optimal resource allocation scheme, which involves parameters such as vehicle cost coefficient, vehicle allocation variable, time cost coefficient, and time adjustment variable. These parameters are used to dynamically adjust resources, such as adding vehicles or adjusting schedules, to offset the impact of events on the route, improve the operational robustness and resource utilization of garbage trucks, and ensure service continuity.

[0090] It should be noted that the elastic scheduling scheme is a dynamically adjusted plan generated through resource scheduling optimization. This means a flexible path and resource allocation scheme, including modified path point sequences, resource allocation details, and contingency measures instructions. These data originate from the output of the optimization algorithm and serve as input for instruction generation and transformation, providing an executable scheduling strategy that enables garbage trucks to adapt to real-time event changes, reduce delays and resource conflicts, thereby improving system adaptability and overall operational efficiency.

[0091] It should be noted that the instruction generation and conversion is based on the elastic scheduling scheme. It is a method that transforms the abstract scheme into machine-readable instructions through an encoding algorithm. Specifically, it includes: encapsulating path point coordinates, time instructions, resource allocation commands, and event response codes using a structured data format; the method involves a serialization step, which maps key parameters in the scheme to standard instruction fields, used to transform the elastic scheduling scheme into a specific executable route instruction set, which is directly integrated into the navigation and resource management system of garbage trucks to achieve automated control and real-time response, thereby improving operational efficiency and accuracy.

[0092] S7. Based on real-time feedback of changes in waste volume and traffic conditions, the route instruction set is dynamically re-optimized and evaluated to generate an updated route for the waste transport vehicle.

[0093] In this embodiment of the invention, the dynamic re-optimization and evaluation of the route instruction set based on real-time feedback of waste volume changes and traffic condition data to generate an updated route for the waste transport vehicle includes: S901 integrates real-time feedback data on changes in waste volume and traffic conditions to obtain a dynamically updated dataset. S902, the dynamically updated dataset and route instruction set are re-optimized and evaluated to obtain the evaluation results; S903, adjust the route based on the evaluation results to generate an updated route for the garbage truck.

[0094] It should be noted that the data fusion operation is based on real-time feedback of waste volume changes and traffic condition data. The data is integrated through a weighted fusion algorithm to generate a dynamically updated dataset. The real-time waste volume change data includes waste volume change values ​​at multiple time points, and the real-time traffic condition data includes traffic condition values ​​at the corresponding time points. The weighted fusion algorithm calculates a dynamically updated data value for each time point. This value is the product of the waste volume change value and the waste volume weight, plus the product of the traffic condition value and the traffic condition weight. The sum of the waste volume weight and the traffic condition weight is 1. The weight values ​​are calibrated based on historical data, with the waste volume weight set to 0.6 and the traffic condition weight set to 0.4 to balance the impact of waste volume and traffic.

[0095] It should be noted that the re-optimization evaluation is a method of re-evaluating routes based on dynamically updated datasets and route instruction sets, using cost-benefit analysis algorithms. The evaluation method includes: first, calculating the expected total cost of the current route instruction set under the dynamically updated dataset, including distance cost, time cost, and resource consumption cost; second, comparing this cost with the baseline cost of historical optimized routes, using threshold judgment; and finally, conducting a feasibility analysis in conjunction with business constraints to identify the shortcomings of the route instruction set under the current real-time conditions, quantify changes in route performance, provide data-driven basis for route adjustments, ensure that the routes of garbage trucks always remain optimal, and improve operational efficiency and adaptability.

[0096] It should be noted that the evaluation results are quantitative outputs generated through re-optimization evaluation. They represent a summary of the applicability of the route instruction set in the current dynamic environment, including: adjustment flags, cost deviation values, adjustment suggestions, and risk levels. These data originate from the output of the cost-benefit analysis algorithm and serve as direct inputs for route adjustments. They clearly indicate whether and how to modify the route, reduce decision-making uncertainty by providing specific adjustment directions, and support fast and accurate path optimization.

[0097] It should be noted that route adjustment is a method of modifying routes based on evaluation results and executed through a path reconstruction algorithm. The adjustment method includes: if the evaluation results indicate that adjustment is needed, a mixed-integer programming solver is used to recalculate the path point sequence, taking into account real-time factors in the dynamically updated dataset, and injecting business constraints and time window constraints. Specific steps involve updating path selection variables and state variables, generating a new ordered path point sequence, which is used to transform the evaluation results into a practically executable updated route. This ensures that the garbage truck routes adapt to real-time changes, minimizes total cost, and mitigates risks, thereby improving transportation efficiency, resource utilization, and system robustness.

[0098] like Figure 2 The diagram shown is a functional block diagram of a route planning system for garbage trucks based on multi-source data, provided in an embodiment of the present invention.

[0099] The multi-source data-based route planning system 100 for garbage trucks described in this invention can be installed in an electronic device. Depending on the functions implemented, the multi-source data-based route planning system 100 may include a data cleaning module 101, an anomaly detection module 102, a garbage source tracing module 103, a target constraint module 104, a parameter optimization module 105, a flexible resource scheduling module 106, and a dynamic re-optimization module 107. The modules described in this invention can also be called units, referring to a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, stored in the memory of the electronic device.

[0100] In this embodiment, the functions of each module / unit are as follows: The data cleaning module generates a baseline route based on historical collection and transportation data. It cleans the pre-acquired garbage volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data to generate a cleaned dataset of garbage trucks. The anomaly detection module performs multi-source temporal feature fusion anomaly detection on the cleaned dataset to generate anomaly detection results for the garbage truck. The waste tracing module performs trajectory inversion on the cleaned dataset to generate the waste tracing results for the waste transport vehicle; The objective constraint module, based on the cleaned dataset, anomaly detection results, waste tracing results, and the baseline route, applies objective constraints to a preset mixed integer programming model to generate an initial optimization model for the waste transport vehicle. The parameter optimization module performs gradient descent parameter optimization on the initial optimization model to generate an optimized path scheme for the garbage truck. The elastic resource scheduling module, based on a large event contingency plan library, performs elastic resource scheduling on the optimized route plan and generates the route instruction set for the garbage truck. The dynamic re-optimization module performs dynamic re-optimization evaluation on the route instruction set based on real-time feedback of changes in waste volume and traffic conditions, and generates updated routes for the waste transport vehicles.

[0101] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0102] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0103] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0104] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0105] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0106] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A route planning method for garbage trucks based on multi-source data, characterized in that, The method includes: S1 generates a baseline route based on historical collection and transportation data, and cleans the pre-acquired garbage volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data to generate a cleaned dataset of garbage trucks. S2, perform multi-source temporal feature fusion anomaly detection on the cleaned dataset to generate the anomaly detection result of the garbage truck; S3, Perform trajectory inversion on the cleaned dataset to generate the waste source tracing results of the waste transport vehicle; S4. Based on the cleaned dataset, anomaly detection results, waste tracing results, and the baseline route, apply objective constraints to the preset mixed integer programming model to generate the initial optimization model of the waste transport vehicle. S5, perform gradient descent parameter optimization on the initial optimization model to generate an optimized path scheme for the garbage truck; S6, Based on a large-scale event contingency plan library, perform flexible resource scheduling on the optimized route plan to generate the route instruction set for the garbage truck; S7. Based on real-time feedback of changes in waste volume and traffic conditions, the route instruction set is dynamically re-optimized and evaluated to generate an updated route for the waste transport vehicle.

2. The route planning method for garbage trucks based on multi-source data as described in claim 1, characterized in that, The process of generating a baseline route based on historical collection and transportation data involves cleaning the pre-acquired waste volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data to generate a cleaned dataset of garbage trucks, including: S201, perform path clustering analysis on historical collection and transportation data to generate baseline routes; S202 integrates the pre-acquired waste volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data to obtain a multi-source raw dataset; S203, imput missing values ​​in the multi-source original dataset to obtain the imputed dataset; S204, perform outlier detection on the filled dataset to obtain a cleaned dataset.

3. The route planning method for garbage trucks based on multi-source data as described in claim 1, characterized in that, The step of performing multi-source temporal feature fusion anomaly detection on the cleaned dataset to generate anomaly detection results for the garbage truck includes: S301, Perform multi-source temporal feature extraction on the cleaned dataset to obtain the multi-source feature vector set of the garbage truck; S302, perform feature fusion on the multi-source feature vector set to obtain the fused feature matrix of the garbage truck; S303, perform anomaly detection analysis on the fused feature matrix to obtain the anomaly score of the garbage truck; S304, perform a threshold judgment on the abnormal score to obtain the abnormal detection result of the garbage truck.

4. The route planning method for garbage trucks based on multi-source data as described in claim 1, characterized in that, The step of performing trajectory inversion on the cleaned dataset to generate the waste source tracing results for the waste transport vehicle includes: S401, Extract the location coordinate sequence and timestamp sequence from the cleaned dataset to obtain the original trajectory data of the garbage truck; S402, Perform trajectory smoothing processing on the original trajectory data to obtain a smooth trajectory sequence of the garbage truck; S403, Perform a stop point detection operation on the smooth trajectory sequence to obtain the stop point set of the garbage truck; S404. Using a density clustering algorithm, the set of stopping points is analyzed to determine the waste source region, thereby obtaining the waste source tracing results of the waste transport vehicle.

5. The route planning method for garbage trucks based on multi-source data as described in claim 1, characterized in that, Based on the cleaned dataset, anomaly detection results, waste tracing results, and the baseline route, an initial optimization model for the waste transport vehicle is generated by applying objective constraints to a pre-defined mixed-integer programming model, including: S501, Based on the cleaned dataset, anomaly detection results, waste tracing results, and the baseline route, extract the optimized parameter set of the waste transport vehicle to obtain the model input parameters; S502, Based on the model input parameters, construct the objective function and constraint set of the mixed integer programming model; S503, apply objective constraints to the objective function and constraint set to obtain the initial optimization model of the garbage truck.

6. The route planning method for garbage trucks based on multi-source data as described in claim 5, characterized in that, The objective function and constraint set for constructing the mixed-integer programming model based on the model input parameters include: S601, Based on the model input parameters, define the set of decision variables for the mixed integer programming model; S602, Based on the set of decision variables and model input parameters, construct the objective function of the garbage truck; S603, Based on the model input parameters and business requirements, define the constraint set for the garbage truck.

7. The route planning method for garbage trucks based on multi-source data as described in claim 1, characterized in that, The step of optimizing the initial optimization model using gradient descent parameters to generate an optimized path scheme for the garbage truck includes: S701, Perform parameter initialization on the initial optimization model to obtain an initial parameter set; S702, perform gradient descent iterative optimization on the initial parameter set to obtain the optimized parameter set; S703, perform path reconstruction on the optimized parameter set to obtain the optimized path scheme for the garbage truck.

8. The route planning method for garbage trucks based on multi-source data as described in claim 1, characterized in that, The method of elastically scheduling resources for the optimized route plan based on a large event contingency plan library, and generating the route instruction set for the garbage truck, includes: S801, perform event matching between the optimized path scheme and the large event contingency plan library to obtain a matching contingency plan set; S802, perform resource scheduling optimization on the matching plan set to obtain an elastic scheduling scheme; S803, perform instruction generation and conversion on the elastic scheduling scheme to generate the route instruction set for the garbage truck.

9. The route planning method for garbage trucks based on multi-source data as described in claim 1, characterized in that, The method of dynamically re-optimizing and evaluating the route instruction set based on real-time feedback of waste volume changes and traffic condition data to generate updated routes for the waste transport vehicles includes: S901 integrates real-time feedback data on changes in waste volume and traffic conditions to obtain a dynamically updated dataset. S902, the dynamically updated dataset and route instruction set are re-optimized and evaluated to obtain the evaluation results; S903, adjust the route based on the evaluation results to generate an updated route for the garbage truck.

10. A route planning system for garbage trucks based on multi-source data, characterized in that, The device includes: The data cleaning module generates a baseline route based on historical collection and transportation data. It cleans the pre-acquired garbage volume data, vehicle status data, real-time traffic data, water and electricity data, and fuel consumption data to generate a cleaned dataset of garbage trucks. The anomaly detection module performs multi-source temporal feature fusion anomaly detection on the cleaned dataset to generate anomaly detection results for the garbage truck. The waste tracing module performs trajectory inversion on the cleaned dataset to generate the waste tracing results for the waste transport vehicle; The objective constraint module, based on the cleaned dataset, anomaly detection results, waste tracing results, and the baseline route, applies objective constraints to a preset mixed integer programming model to generate an initial optimization model for the waste transport vehicle. The parameter optimization module performs gradient descent parameter optimization on the initial optimization model to generate an optimized path scheme for the garbage truck. The elastic resource scheduling module, based on a large event contingency plan library, performs elastic resource scheduling on the optimized route plan and generates the route instruction set for the garbage truck. The dynamic re-optimization module performs dynamic re-optimization evaluation on the route instruction set based on real-time feedback of changes in waste volume and traffic conditions, and generates updated routes for the waste transport vehicles.

Images